STOP
Early Journal Content on JSTOR, Free to Anyone in the World
This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in
the world by JSTOR.
Known as the Early Journal Content, this set of works include research articles, news, letters, and other
writings published in more than 200 of the oldest leading academic journals. The works date from the
mid-seventeenth to the early twentieth centuries.
We encourage people to read and share the Early Journal Content openly and to tell others that this
resource exists. People may post this content online or redistribute in any way for non-commercial
purposes.
Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early-
journal-content .
JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people
discover, use, and build upon a wide range of content through a powerful research and teaching
platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit
organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please
contact support@jstor.org.
III.— CONFLATE READINGS OF THE NEW
TESTAMENT.
It has for some time been my conviction that many of the textual
obscurities in the New Testament and other early books are due
to the operation of unrecognized causes, for the want of a perception
of which the scholar is often presented with a text which might,
indeed, be genealogically nearer to the true reading than other
more popular presentations of an author's words, but which, as
regards good sense, are often very widely remote from the truth.
One such hitherto occult factor (I call it occult since I cannot find
any recognition of it in any book that I know) I believe to lie in
the existence of lateral aberrations from one point to a corresponding
point in the columns of an ancient MS, and there are instances to
be found of the application of this remark to textual criticism in
various errors which are alluded to in the supplement to the
1 2th number of the American Journal of Philology. Since
publishing these I have from time to time returned to the
point, and last year endeavored to prepare, by means of these
lateral errors, a complete restoration of the text of the Epistle
of James, line for line and page for page, to its archaic form.
Unfortunately, I have not as yet been able to lay this restoration
before those who are interested enough in New Testament studies
to give the matter a careful examination.
It will easily be seen that any attempt to find the origin of the
textual errors of the New Testament in the early forms of the text
itself, is part of a larger theory, viz. that the study of transcriptional
errors belongs to the paleographer first and the general critic
afterwards. Unhappily, the course of modern criticism, with a few
honorable exceptions, has been in the opposite direction. We often
hear from the critics explanations of scribes' blunders ; their scribe,
however, is not a machine, but a highly cultured cleric, with an
unlimited facility for confusing the Bible with itself and correcting
the New Testament by means of obscure verses in the book of Job,
or Proverbs, or Leviticus. Neither Burgon nor Hort could sit
down to copy the New Testament and make some of the mistakes
which they attribute to the scribes.
On the other hand, when we treat the scribe as a machine
imperfectly adjusted, and examine the errors which he is most
26 AMERICAN JOURNAL OF PHILOLOGY.
likely to make, we find, upon examination, that these are often the
very mistakes which he has made ; and that a very simple
explanation will often replace some obscure assimilation. Even
when a scribe assimilates his text to some other, there is often a
reason to be given for his error, which removes it from the catalogue
of things purely arbitrary. For instance, if we observe that the
structure of all known MSS is rectangular, the principal motions
of a scribe's eye are mechanical motions right and left, and up and
down. It follows, therefore, that the machine-errors to which he is
most liable are right and left errors, and up and down errors. The
latter causes line-omissions and line-repetitions, the former causes
page-aberrations, and the omission or repetition of equal large por-
tions of the text. No one doubts the existence of the vertical error
which is patent in the omission in almost every written document
of lines of the copy ; but a great many people fail to see that the
cause which produces this error can be turned through a right
angle and still retain its efficiency. The reason of this is, that in
ordinary printed texts it is easy to detect the line error, but the
page-error is often veiled, especially in those cases where it is
confined to the borrowing of a few letters or a single word from a
distant part of the copy. Moreover, our modern printed books
seldom show more than two columns to the eye at once, and do
not therefore suggest such errors. A transposed page is, however,
easy to recognize ; this form of mistake is more often due to the
bookbinder than to the scribe.
It must not be supposed, from what has been stated, that the
recognition of the line-aberration has been accompanied by a
proper critical application of the principle of such aberrations to
the text of the New Testament. Some of the best printed editions
are disfigured by ghastly line-errors. For example, Westcott and
Hort print on the very first page of their N. T., Matt, i 7 :
'A/3ia #€ €yevvT)&€i/ rbv Atra^,
'A<ra<f> Se iytvvrjcrev tov 'la>a afjyaT,
laxTCMpaT Se xri
where the reading 'Ao-d<fi arises simply from the corresponding
letters in the word 'ia>o-a<f>d.T. Perhaps an exactly similar explana-
tion holds on the next page, where we read :
Mavacrcrrjs 8e £yfvvr)<rev tov 'Afxws
'A[ia>s 8f iy£w7]<rtv tov laxreiav
l<oo~€ias 5e KT€
CONFLATE READINGS OF THE NEW TESTAMENT. 2"J
It can hardly be accidental that this coincidence of letters is found
in the proper names. And this simple paleographic explanation
being given, is not to be shaken by an array of excellent MSS in
which the archaic error may be preserved. For another instance
take Luke xv 29 ; if the text were written in the form
eACOKd,Ceplct>0N!Nd,
M6TdT0ONc(>IACONW\OY
it would be likely that any transcriber who had written iptywv for
epupov had been guilty of an eye-aberration into the <j>i\av of the
next line ; and then the variant would hardly have been promoted
to the margin of Westcott and Hort's text on the single authority
of Cod. B.
Nor should we, in Mark vi 14, read fXcyov for e\eyev, if we
observed that in
TOONOMAAYTOYKAie
AeroN
the tense has been affected by the first syllable of oVo^a. Neither
should we read liriyvao-av in Mark vi 33 for iyva>o-av if the text were
printed
erNCOCANTTOAAOIKAl
rrezH
where the inserted letters are evidendy taken from the line below.
We might say much more on the subject of line-errors, but for
the present let this suffice. We pass on to make a few remarks
on the page-errors.
Let us then once more demonstrate the frequency of the error
to which we give the name of lateral aberration, or parablepsy.
The importance of the study of the lateral aberrations becomes
more and more evident as we examine microscopically the text of
the New Testament ; it is probable that not a single one of the books
of the New Testament, except some of the shorter epistles, has
escaped from this error. And in many cases the most perplexing
variants are cleared up by the recognition of the existence of such a
source of error in columnar texts. We shall give a string of
illustrations. For example, in the text of Matt, xiii 35 we have a
very difficult reading in which there is the following distribution
Of authority : ojrwr ir\i]pa>6rj to prjde v Sia tov wpoKprjrov Xeyovros, 'Avoiga
iv irapa^oKais to aTopa p.ov ktc by almost all authorities. After fita is
28 AMERICAN JOURNAL OF PHILOLOGY.
added 'Hmu'oi by H* i, 13-124-346, 33, 253 rushw. aeth. cod.,
Horn. CI. Porphyr. (cf. Brev. Psalt. in Hier. Opp. vii 270, Vail.) 1
If we were left simply to balance these authorities, we should,
I think, conclude in favor of the ordinary reading. But the affair
is not so simple as at first sight appears. For, as pointed out by
Dr. Hort, we have the evidence of Jerome that there was a third
reading, in which 'Ao-d0 takes the place of 'Ha-aiov. And although
this reading is not to be found in any existing MS, yet according
to the Brev. in Psalmos, we are assured that it was found in all
old MSS, but was removed by ignorant men ; that by an error of
scribes 'Ha-aiov was written for 'Ao-a$, and that at the time of writing
many copies of the Gospel still had 'Ho-aiov.
I propose to show that 'Aadcj> may be the correct reading,
although it has disappeared from all known MSS. In order to
see this, let us write the passage in question out conjecturally in
the style of an early codex or paper roll.
Matt, xiii 35 : oncocnAHpco9H ma9ht<m&ytoy
TOpH6eNAIAC*4> AerONTeCAIACd(|)
Toyrrpo(|3HTOYAe hcon hmin . . .
P0NT0CAN0I5CO
eNTT&p&BOA&ICTO
ctomamoy epey
50M<MK6KpYMMe
NAATTOK&T&BOAHC
T0T6d(t>eiCT0YC0
XAOYCHAeeNeic
OIKI&NKAITTpOC
HA9ANAYT000I
A single glance at the second lines of the two columns will suggest
that six letters in one column were transferred from the other ;
and whether we have hit upon the exact arrangement of text or
not, the concurrence is difficult to explain except by some such
restoration. We must now take one of two hypotheses : (a)
Column 1 took the letters from col. 2 ; (0) The converse.
In discussing the first, we are to remark that in the assumed case
we have to replace a right reading (whatever it may be) by a
wrong one which itself is so apposite to the case that it would be a
perfect marvel in the history of curious coincidences. Surely no
1 Hort, Notes on Select Readings, p. 13.
CONFLATE READINGS OF THE NEW TESTAMENT. 29
scribe remembered that Ps. LXXVIII was attributed to Asaph
when he saw the words in an adjacent page. And if we suppose the
original reading to be 'Ha-awv we have to assume that a happy-
accident corrected into sense an absolute falsity, and removed a
reading the acceptance of which strains all our faith in the accuracy
of the writer who primitively set it down.
In the other hypothesis there is no difficulty : for not only have
we very good patristic authority for the existence of the variant,
but we have the additional confirmation that the displaced reading
of the second column has been preserved to us. If there were no
variant in the second column we might, perhaps, feel a residual
hesitancy ; but a reference to the textual apparatus in any New
Testament will show that the proper reading is not 8iao-d<pr)o-ov but
(ppuaov. For example, the following is the note of Tregelles.
Siao-a^o-ov B. Orig. iv 254" enarra. a. b. g l k (narra^" 2 ) | cppaa-ov St.
CD. rel. Orig. iii 3". 4"°, 442", 481", edissere (diss.) Vulg. CI (Am)
( c )f- (ff^g* (vid. cap. xv. 15). To which it must be added that
the Sinaitic Codex has Biaa-cKprja-ov corrected to <ppd<rov by an early
hand. If, then, we were reasoning simply upon the grouping of
the MSS, we should, perhaps, incline with Tregelles and Westcott
and Hort to follow the concurrence of KB and early Latin copies ;
but the cause of the variant being known, the variant, however
ancient it may be, must disappear. Early the error must, of course,
be, as indeed are all important errors ; but this has an especial
antiquity, since it has affected copies which between them can
produce a singularly pure text.
And now let us gather up the conclusions which would follow
from our conjecture.
1. The primitive page in Matthew, or at least a very early
page, from a copy which is genealogically ancestral to almost all
our early texts, contained about 160 letters.
2. Since the Sinaitic Codex has preserved the lateral error, while
writing ^a-alov for d<rd<p, we may endorse the statement of the
Breviarium in Psalmos as to the order of genesis of the separate
errors.
3. We note that Tregelles, and, no doubt, other writers, have
assumed that <ppda-ov is an assimilation of text to the 15th verse of
the XVth chapter. This exaggerated doctrine of assimilation has
led to frequent errors. The present case is one in point.
4. We cannot close our remarks without adding that here we
have an instance which we believe a closer scrutiny would parallel
30 AMERICAN JOURNAL OF PHIL0LG0Y.
from many other obscure passages in the Scripture, of an original
reading no longer extant.
As soon as we have recognized this archaic arrangement of the
pages we have a shrewd suspicion that the very same MS or
series of MSS in which the error already alluded to was made and
preserved, is responsible for an omission in Matthew xvi 2, 3,
'otyias yevofihtjs Acyere kt4. For on examining the passage it is found
to contain 162 letters, which agrees very closely with the previous
estimate. Dr. Hort says of this passage, that " both documentary
evidence, and the impossibility of accounting for omission, prove
these words to be no part of the text of N. T." The omission of a
single page is a sufficient explanation, and in nowise affects the
documentary evidence in other passages, from which an induction
has been made to the case under consideration.
A second instance shall be given from the same Gospel. The
error, again, is one which has altogether disappeared from the
copies. In the time of Origen, however, we find that in the xxi
of Matthew, the expression 'Qo-awa tw vly AaveiS was changed into
'Qa-awa ra oiKa> Aavei'S, in one of the two verses (xxi 9, 15) where
the words occur. Origen will have vl<$ read in both places. 1 An
examination shows that the error was made in the 13th verse by
lateral aberration from the 15th, probably in some copy immediately
ancestral to Cod. B. For in this MS we have in the 34th line of
the first column of a page
TTTAIOOIKOCMOYOIKOC
and in the fifth line of the second column,
fO NTACCOCAN N ATCO Y
ICO
the interval between these lines being 14 of the lines of B or one-
third of its columns.
As an illustration of the same mistake in the Gospel of Mark we
may take Mark v 1, where by borrowing four letters from the
fifth verse, the reading Ta8apt]»S>v has been changed by means of
the word lj/wpas lying parallel with it, into Tepaar)v5>v. It is possible
that this explanation may be thought fanciful ; stranger ones will
be given presently and with greater certainty. From the Gospel
of Luke a remarkable case will be given later on.
1 See Tischendorf, in be.
CONFLATE READINGS OF THE NEW TESTAMENT. 3 1
In John v 37 we have the following arrangement from Cod. B.
Te(J3coNHNAYToynco
TTOTeAK HK0AT60YT6
eiAocdYToyeoopAKATe
Kdl TONAOfONAYTOY
oYKexeTeeNYMiNwie
N0NTA0TI0NAneCT6l
AeN6KeiN0CT0YTC0
YwieicoY nicTeYETe
epdYNATeTAcrpAtt^c
OTiYMeiciOKeueeN
dYTAIC ZCOHNAICONIO
exeiN KAI 6K6INAI 6ICIN
Line 41, rr<s,THpei<eiNOC iwewi^p aimaptypoycai nepi
42, TYpHKeNnepieMOYOY eiv\OY KdioYOeAeTeeA. Line 14
The influence of the second page on the first has produced
eK€~ivos in the first line for alros. The two readings are undoubtedly-
early, since they are conflated in Cod. D into i kcivos airos " (perhaps
eiceivos corrected over line to airds). I suppose we must call the
reading airos Western and Syrian. In spite of this it seems to
be correct.
As an illustration from the Pauline Epistles we may take 1 Cor.
ix 9, where ol <£i/ia><r«s @ovv ako&vTa is by some copies corrected to
oi Ktjfiaxrus kt{, under the influence of k^Hs in v. 5, where the
syllable kj] may have been the last syllable in a line, as it is in Cod.
B to this day.
From the Catholic Epistles we give the following from James
(in which epistle there are at least five). The instance is taken
from the pages as we have tried to restore them :
p. 33, line 9, ky lAOYorecoproceK
10, AexeTAI TON TIMION
11, KdprrON
p. 34, line 9, MBeTeAA6A4x>iTHC
10, MAKpOGYMIACTOYCTTpO
11, CJ5HTAC
In this passage (lac. v 10), Cod. 13 by aberration reads after
naKpoOvfiias the impossible Severe, which has been softened to exert
by A. H°. 5.40 mg. 73, and the Ethiopic version.
In the Epistle of Jude, short as it is, there is a curious aberration
which has much confused the text ; the word &na% of the 3d verse
32 AMERICAN JOURNAL OF PHILOLOGY.
having crept into the beginning of the fifth. We have no room to
give the pages in full.
It must be admitted that these errors afford us great assistance
in the restoration of the early text-forms, and in the intimately
connected problem of the genealogy of the witnesses that remain
to us. We shall show first the way in which early forms may, by
the means indicated, be theoretically restored.
Let us then, in the first place, imagine to ourselves a MS written
uniformly, so that there are m letters to a line, and n lines to a
page. And suppose that after a certain number of transcriptions
in this form a new pattern of script is introduced, consisting of p
letters to the line, and q lines to the page ; and so on continually,
the forms being allowed to persist longer in the earlier copies than
in the later ones, although this is not a necessary part of the
following argument. Then let us ask ourselves what kinds of
errors are most likely to present themselves in the successive and
final texts, assuming them to be copied by scribes of average
carefulness. In the first place, we notice that the first copy made
being liable to the eye-error of vertical aberration (generally
induced by similarity of the letters in separate lines) will exhibit a
majority of errors, which are either m letters in length or a
multiple of m, such as 2m, yn, etc. The ^-errors will, however,
by far predominate. And each successive transcription will cause
these errors to accumulate, until after a little time the mere
registration of the variants would be sufficient to indicate the
original form of the text, even if that form should be deserted.
And now let the form be changed to that indicated by the letters
p, q. The same thing will occur here, and we shall have an
accumulation of/-errors, which, like the former ones, soon become
by their multiplicity self-betraying as to the form of text in which
they were made. And so we might carry the matter forward.
Finally, if we denote the MSS which have adhered to the first
form by the letters Mi, Ma . . . , those which have adhered to the
second form by Pi, Pa . . . , to the third by Si, Sa . . . and so on,
we shall have the following results upon classifying the variants of
all existing copies, viz. the insertions in or omissions from a
standard text :
(i) A number of copies will exhibit, among the variants, a
preference for variants of a given length, which is found to be m-
letters.
(2) Another group of copies will exhibit a preference for errors
of two given lengths, viz. m, p.
CONFLATE READINGS OF THE NEW TESTAMENT. 33
(3) A third group will suggest errors of lengths of m, p, s. And
so on, the phenomena rapidly tending to obscure one another.
In determining which of these errors, from a standard text, to
reject or accept, we must remark as follows : it is perfectly easy to
omit a line of a text in copying it ; but to add foreign matter to it,
which shall precisely be equivalent to a line of the text, may be
assumed to be very unlikely ; and, therefore, the majority of a
group of equivalent errors whose length is equal to a line of the
text are pure omissions. The only case in which this breaks
down will be the following : Suppose that a given line of a copy
has been affected by some scribe's stupidity, so as materially to
change the sense without affecting the length (as by the substitution
of two or three letters from a wrong line), and that by the subsequent
correction of the passage two readings have been placed in close
relation, it frequently happens that the real line and the erroneous
line which is equal in length to it, both combine to form a new
reading, which has thus increased the text by one of its own lines.
This phenomenon is known by the name of conflation. Setting
aside the phenomenon of conflation, then, we say that the line-errors
of a codex to which it shows a peculiar liability are omissions, and
not additions. In dealing, therefore, with our groups of MSS, we
must first restore to the texts denoted by the letter M, their w-line
omissions ; similarly with the P-texts and the S-texts, etc. As
soon as the texts come to be broken up into a non-uniform script,
the above reasoning fails, except in so far as it shows original
errors conserved from the various forms through which the text
has passed.
The relative antiquity of the texts, supposed uncomplicated by
mixture, will then appear at once by the consideration that the
M -texts have no ^-errors, while the P-texts show both /-errors and
»«-errors, and, therefore, in the discussion of any given error, the
M-texts have the greater weight, except in the discussion of w-errors.
We must now return to the original text, and remark that not only
is such a text, when copied, liable to line-aberrations, w-letters, 2m,
etc. ; but there is a danger of aberration from column to column,
or from one column of writing to another two columns distant.
These errors,which are far more frequent than is generally supposed,
will, when they can be recognized, supply intervals whose average
is mn-letters, or 2mn, etc. ; and by means of these errors we can
restore the original page ; mn being given by the lateral aberration
and in by the average vertical aberration.
A succession of copies will now exhibit as follows : the copies
34 AMERICAN JOURNAL OF PHILOLOGY.
marked M will exhibit long aberrations (in which must also be
included transpositions of early pages) of an average length mn :
the P-copies will exhibit average errors pq as well as by inheritance,
errors of length mn : and so on. By repeating the error of
aberration, whose measure is mn, the P-copies again demonstrate
the M-type to be the earlier, and in this way a firm grasp may be
obtained over the genealogy of the group of MSS which are
placed under our consideration.
The advantage of this method is that it is purely scientific : a
careful observation, for instance, will show that the New Testament
documents grow smaller and smaller, both as to lines and pages,
as we come nearer to the first centuries ; and, therefore, the more
nearly do they of necessity approach to fixed types of writing, as
to length of line and page. The margin of variation of size being
thus diminished, it would be possible to pick out the earliest sizes
by the aid of the earlier errors, even if there were not, as we
believe there is, a peculiar reason for the adoption of lines of given
length. Moreover, we must not forget that, however little modern
documents seem to lend themselves to the theory of aberration by
lateral error, the case is widely different when the columns are
narrow, and when, as in a papyrus document, many of them may
be under the eye at once. Assuming, then, the existence of the
vertical and lateral aberrations, we proceed to apply our argument
practically to the determination of the texts and text-forms of
the New Testament. We begin as follows :
Conflate Readings.
From what has been already said it appears important that, as
far as possible, we should remove from the examination those
phenomena which are due to conflation of separate readings. If
we can do this in any particular book, or MS of that book, we
shall have taken at once a step in the classification of the MSS
which furnish the contending members of the conflation, and in
many cases we shall obtain a clue to the original structure of the
copy in which the conflation occurs.
It is well known that the most powerful part of Dr. Hort's great
Introduction to the New Testament consists in the exposition of
eight cases of conflation in the early texts of Mark and Luke.
Nothing has been more fiercely assailed, nor as yet with less success,
than this stronghold of the new textual system. All that Burgon
and Cook have been able to do in attempting to demolish the
cumulative argument of Dr. Hort, by the denying each instance
CONFLATE READINGS OF THE NEW TESTAMENT. 35
of conflation in detail, has been the maximum of effort and the
minimum of result. For, it may be observed, it is not necessary
to Dr. Hoit's theory that the whole of these eight conflations
should be verified ; a single one, correct in all its details, would
prove the chronological subordination of the texts which give
combined readings to those which do not make the combination.
And, moreover, there are certain considerations which present
themselves at once to an enquiring mind : granted that there are
groups of manuscripts say P, Q, R, which exhibit the peculiarity
that R has readings combined out of separate readings in P and Q,
so that we are shut up to the hypothesis either of a conflation on
the part of R, or of two separate alternative omissions on the
part of P and Q, surely the quicker way to upset the conflation
hypothesis would be to bring forward some case in which such a
group as Q had united readings out of P and R. But this has
not been done, for the simple reason that no such phenomena are
forthcoming. If they were, would they not be a reductio ad
absurdum for the theory of conflation ?
The fact of the matter is that the conflations cannot be wholly
denied ; and the conclusions which follow almost as a matter of
course, from their admission, can only be evaded by a more careful
examination of the argument, especially of one point, which both
Dr. Burgon and Dr. Hort seem to me to have unhappily missed.
It is well known that after Dr. Hort has divided his authorities
into the three camps which he designates by the names Neutral,
Western, and Syrian, that he rejects the Syrian readings on the
ground of conflation, and the Western readings on account of
multitudinous eccentricities to which the texts that contain them
are liable. It is this last step to which I object. I agree to one
group of witnesses being rejected or undervalued for proved
or probable fabrication of text, but think the other group has
been unduly depreciated. And in order to settle the ques-
tion for myself, I have tried to go more closely into the case,
especially in the following points: (i) Why are all the typical
conflations in Mark and Luke, and none out of Matthew ? (ii)
Whence did the separate members of the conflated text arise, since
both of them by hypothesis cannot be original ?
To the first of these questions I have not been able to give an
answer, however convinced I may be by the law of probabilities
as to the antecedent unlikeliness of the existing circumstances. To
the second, however, I find myself able to reply in some measure.
And perhaps my explanation will not only throw some light on
36 AMERICAN JOURNAL OF PHILOLOGY.
the history of the text, but will be an illustration of the great
canon of criticism, of which all the others are only uncertain and
variable expressions, that when the cause of a variant is known the
variant itself disappears.
The first thing, then, that we notice about the texts which are
said to be conflated is the fundamental equality which generally
prevails in the separate members.
Taking, for instance, the simplified texts, to which we are led by
the analysis of Dr. Hort, we find, Mark vi 33, the readings rat
Trporjkdov airols, (ecu avvrjkdov alrov. The first of these readings is
17 letters, the second is 16.
The second passage is Mark viii 26. It turns upon the variants
Mrjde els Trjv kwiijjv (laeKdrjS,
and Mi/Sfvt etirgs els rrjv Ka>/trjv.
Of these the first part is 23 letters, and the second 22.
The third passage is Mark ix 38 : the alternative members of
which are rat eKoKio/iev avrov, on ovk aKoKnvdel rjfxiv and or ovk dxoXov&t
lied' fi/imv, ral fVwXvo/uev oOtov. The first of these is 36 letters, the
second is 38.
The fourth passage is more difficult to handle critically, and as I
do not want to force the texts to prove a theory, I omit it.
It appears, therefore, that of four passages selected to illustrate
conflation in the Gospel of Mark, three exhibit an almost exact
equality of the separate members. (I omit those in Luke for the
present.) Now I think it will be admitted that it is not a mere
chance that this equality subsists ; upon the theory which asserts
conflation it becomes almost inexplicable that this peculiarity should
be so prominent, unless it be admitted that one of the separate
members is an early and slight distortion of the other ; an assump-
tion which is almost implied in the theory of a single ancient text,
from which all existing texts have been derived.
And so we come back to the question, how did one element of a
conflate text arise out of the other ? — a point upon which I have
meditated often and long ; sometimes proposing to myself the
hypothesis of different translations of an early Aramaic text and
sometimes giving the thing up as an insoluble riddle. However,
the fundamental feature of the passages examined being their
equality, it seems most reasonable to go back and seek the origin
of the various readings in the simplest solution, a scribe's blunder.
Let us examine Mark vi 33. We have, writing the verse out,
CONFLATE READINGS OF THE NEW TESTAMENT. 37
KAieiAONAYTOYCYTTA
!~0NTACKAierNCOCAN
TTOAAOIKAITTeZHATTO
TTACOONTCONTTOAeOONKAl
CYNeipAW\ON6K6IKAI
np0HA60NAYT0YCKTe'
It will be observed, on writing the text in this way, that the
concurrence of ku! at the end of two successive lines invites the
vertical error aw for npo, and produces at once the necessary second
variant, from which all the rest, and the conflated texts, can be
derived. The error is, therefore, at once explained by the assumption
of an early text written in lines of 16-18 letters each. It is this form
which has survived with little change in Codex B, which has pre-
served the correct reading, and thus vindicates itself as an early
text, more ancient than those of the same type as itself in which a
particular error was made. And, on the other hand, the Western
texts are shown to bear witness that this is one of the primitive
forms of the text, by the fact that they have preserved the error
made in that form. On both grounds Dr. Hort's conclusion as to
the superiority of the text of B is in this case confirmed. As soon
as this instance has been settled, we see the reason of the third
selected conflation, Mark ix 38. We have only to write the text
as follows :
6KBaAA0NTAAAIW\0NIA
OCOYKAK0AOY6eiHMl
KAieKCOAYOMgNAYTO
OTIOYKAKOAOYBeiHwil
to see that in a text, written 18 letters to the line, a line has been
accidentally repeated, or has first been moved over another line,
and then conflated into a new reading. But in this case we cannot
affirm with certainty which of the two separate readings is the
earlier. All that we can say is that the text which lends itself most
readily to the production of error is one written in lines of 18 letters.
But it lends itself almost as readily to the production of error on
the hypothesis that the longer recension is the correct one. We
must not be positive that because Cod. B has followed very nearly
the 18-line type, therefore its reading is the more ancient and
correct one, for the omission of 18 letters is just the error to which
that text is liable.
The fourth conflation may be explained in the same way. The
text in Mark ix 49 may be conjectured to have stood as follows :
38 AMERICAN JOURNAL OF PHILOLOGY.
KAITO
n ypoycBe n n YTAirr dCA
eyciAAAiAMcBHceTAi
and by an error of the scribe three letters from a preceding line
were repeated, making mpl, and the final a of Bvaia was immediately
absorbed in the two similar succeeding letters. From this the
repeated a\i was dropped, and a connecting particle introduced by
some MSS, although we find oXi retained after nvpi by Cod. X.
Cod. K is taken from a Greek exemplar which read nao-a (? yap)
dvma avaXmBrjcrerai, which is, as Dr. Hort points out, a corruption of
the preceding.
In this case then, the judgment seems to be in favor of the
Western MSS, and the far-fetched reference to the book of Leviticus
as an inducing cause of error may be rejected. The error being of
the length of 18 letters may be corrected without severely lowering
the high estimate we have of the value of the neutral text. We have
now discussed all except one of the conflations cited from the
Gospel of Mark, and shall be able to do something presently
towards the investigation of the genealogical relations of the
documents. We will leave the other passage for the present.
All that we have shown thus far is the existence of early 1 8-line
texts of Mark which are a sufficient explanation of several
important errors. Let us now go on to examine a passage in Luke
which will illustrate the doctrine of the lateral aberrations.
The eighth conflation of Dr. Hort occurs in the last verse of
Luke, in which we have the two readings ahovvra tov 6c6v and
cvXoyovvres rbv 6t6v. We seek to explain one reading by the other.
Let us write out at length the last three verses of Luke as they
stand at the close of the gospel in the Codex Sinaiticus :
Col. 4.
Line 13, KdiereNeToeN
TCoeYAO.riN&YTO
AYTOYCAieCTH
attaytconkaiay
TOITTpOCKYNHCA
T6CAYT0N YTT6C
Tpe^ANToeiciepoY
caahm'mgtaxa
pACMer^AHCKAl
Col. 4. HCAN AIATTANTOC
Line 11, tacxipacaytoyh eNTcoiepcoeyAo
12, YAOfHCeNAYTOYC rOYNTeCTONSN
CONFLATE READINGS OF THE NEW TESTAMENT. 39
Here we find that the eJXoyouvres has come in, by lateral aberration
to the last page from the last page but one. The explanation is
perfect. (Tischendorf saw that this word was the cause of the
error and corrected his text accordingly against the Sinaitic
Codex.) We infer: (1) that the early page of Luke was not
substantially different from a quarter of the column of the Codex
Sinaiticus ; (2) that in this case the neutral texts must be subordi-
nated to the Western texts, which have preserved the correct
reading ; (3) there is a probability that the archaic line was also
not very different to that of the Sinaiticus. We need hardly say
that this result will very much reverse the method in which the
materials of the text are handled in the Gospel of Luke. It will
enhance the authority of the Sinaitic text in Luke, though not to
the same extent as if the MS had avoided the error.
From this point we can go on to discuss the seventh conflation,
Luke xii 18, in which we are confronted with the four readings:
B T L X mem. 346) tonciton[moy] ? n letters,
a. c. d. e. m. TOYCKdpTTOYCMOY 14 "
(X D. b. ff. r. q and all Syrian texts) T&reNHMATAMOY J 3 "
B. T. L. X. mem. and all Syrian texts) K&iTA&rAOAMOY 13 "
It is needless to say that we cannot take all these readings : we
remark that they are almost all line-lengths, according to the
previous hypotheses. Giving especial weight to the actual reading
of X, we feel tolerably sure that to yevrjuard uov is a line of text. In
the next place we remark that the first line is not complete without
a icat. For it is certainly either a line dropped, or one substituted
for another line of known length, 12-14 letters. The second line
is an assimilation to what has just been written, while rk dyadd looks
like an anticipation of 7roX\a dyadd immediately following. Suppose,
then, we read :
roil trtrov fiov Kai ret yfvfjfiard fiov
(which is preserved only in Cod. 346).
Now the fact that mil was necessary to the completeness of the
first line, shows it to be a genuine part of the text, and not an
alternative line hooked on to another of the same length. Moreover,
of the four MSS 13. 69. 124. 346, the first three read t6v o-It6v fiov «al
to dyadd /xov, while the fourth reads tov oirov fiov Kal to yfvrjfiaTa /xov.
This shows that ra yivrffiard fiov and ra dyadd /iov are alternative
readings, of which the former has evidently the preference for us,
40 AMERICAN JOURNAL OF PHILOLOGY.
as it had for the scribe-corrector in 346. The conflation then
consists in the union of the last two readings by the Syrian text,
while at the same time the first limb of the sentence is dropped.
The discussion of those conflate readings which we are able to
interpret, with any degree of certainty, leads us to the conclusion
that it is impossible to predict correct readings as infallibly belonging
to either group of manuscripts representing the relatively simple
readings. The Western readings are found to vindicate for them-
selves a purity and antiquity which is, in certain cases, greater than
that of the neutral readings. They cannot, therefore, be wholly
rejected or used in the supplementary manner in which Westcott and
Hort employ them. A reading is not to be rejected as Western and
Syrian merely because it is Western and Syrian, for either the
probabilities against such readings have been overestimated or the
results of the textual examination have been too hastily generalized.
The peculiar character of the Western text can, moreover, be
eliminated to a certain extent, by remarking that its errors are those
which are incidental to rapid transcription, and the causes of the
separate mistakes can often be detected. For instance, in Mark
xv 34, the reading of certain Western texts is avelSia-as for e'yKure-
\nres. Now this reading might plausibly be claimed as a modification
of the more difficult thought of divine desertion, or as an assimilation
to other passages of the quoted Psalm. But it is more likely
only an aberration to the &veibi(ov avrov of the 32d verse, the
interval between the confused words being 161 letters. When we
have recognized the error in the Western text as a simple
transcriptional blunder, how does the discovery prove license and
corruption in the remaining Western readings ? And if a number
of them can be explained in this innocent manner, will not our
estimate of the Western readings generally change ? At least the
margin of wilful and artificial change will be much reduced. We
hope to have more to say on this important subject at no distant
date.
J. Rendel Harris.